A reasonable allowed change in tolerance

Main Page | Report this Page

Science Forum Index » Optics Forum » A reasonable allowed change in tolerance

Page 1 of 1

Author

Message

Guest

Posted: Tue Apr 29, 2008 7:12 pm

Hi y'all

What is a reasonable allowed change for a tolerance in wave RMS when
tolerance a system in Code V?

Is .0001 out of the question?

Thanks.

Helpful person

Posted: Thu May 01, 2008 8:47 am

Joined: 22 Jun 2004 Posts: 658

On May 1, 1:16 pm, "Richard F.L.R.Snashall" <rf...@notnotrcn.com>
wrote:

Quote:

Philli...@yahoo.com wrote:
Hi y'all

What is a reasonable allowed change for a tolerance in wave RMS when
tolerance a system in Code V?

Is .0001 out of the question?

Thanks.

Why would a percentage not fit?

It all depends on many different items. In my opinion, tolerancing
(and mechanical mounting) are the most difficult, least taught and
most important items of lens design. It is too complex to answer in
this forum, however here is a list of points to consider to stimulate
your thoughts.

1. Is this a one off or a high volume production or somewhere in
between? Remember, tolerancing is a statistical process.

2. What is your nominal and required performance? If your system is
diffraction limited errrors introduced from various elements do not
add up in a linear manner. In other words, if you plot resolution
against wavefront aberration it is not a linear relationship.

3. How do you intend to mount your lens? Tolerancing cannot be
comnpleted before this is known.

4. How accurately do you know the dimensions of your parts? Instead
of relying on the drawing tolerances you can measure the parts and
adjust the design. For well corrected lenses this is often done with
glass data.

5. Remember that vendors will not usually supply parts with a
gaussian distribution about the nominal dimensions. For example, if
they do not need to rework lenses in their processes then all the
lenses are likely to be at the large end of the tolerances. (Room
left in case rework is necessary.)

6. Take care to include plus and minus field angles and full pupils
when performing tolerance analysis.

7. There is always a cost preformance trade off. This needs to be
considered not only when performing tolerance analysys but also in
deciding which lens construction to use. Certain lens designs are
extremely difficult to manufacture.

These are just a few points.

Has anyone written a book covering "real world" lens tolerancing?

Richard F.L.R.Snashall

Posted: Thu May 01, 2008 12:16 pm

Guest

PhillipKP@yahoo.com wrote:

Quote:

Hi y'all

What is a reasonable allowed change for a tolerance in wave RMS when
tolerance a system in Code V?

Is .0001 out of the question?

Thanks.

Why would a percentage not fit?

Brian

Posted: Thu May 01, 2008 11:37 pm

Guest

On May 1, 7:47 pm, Helpful person <rrl...@yahoo.com> wrote:
<snip>

Quote:

Has anyone written a book covering "real world" lens tolerancing?

Not that I know of, but a chap I worked with at Rank/Marconi/BAe
called Alan Stuart did an MSc thesis on it at Imperial College in the
70s. As he already had years of experience in the industry it made the
document that much more valuable. In those days we had to do all the
sums on a Marchmont calculator, so we really had to understand the
problem before committing to calculation! Later, Harold Hopkins even
produced a brilliant scheme for calculating all the effects of tilt/
decentre tolerances (including three types of distortion) using just
two paraxial rays, like the Seidels. When I worked at Dallmeyer we
even used a Seidel approximation to estimate MTF degradation.

My experience with tolerances was as follows:

PHASE 1

1. Decide an overall permitted degradation (delta MTF or whatever - it
may include more than one criterion)

2. Calculate the sensitivity table for each error (keep symmetrical
and non-symmetrical errors in separate budgets). This is the bit that
used to be a nightmare in the old days, now you can even do it on OSLO-
EDU!

3. Split the degradation between symmetrical and non-symmetrical on an
arithmetic (worst-case) basis.

4. For each error, allocate the budgets on a root-sum-of-squares basis
- or, in other words, for N errors in the budget, multiply each worst-
case tolerance by sqrt(N).

5. Find out the smallest tolerance achievable in the workshop. And
decide on a sensible upper limit for each.

6. Re-distribute the tolerances (you'll probably need to repeat steps
3 to 5 a few times).

7. Round the tolerances off and check the total RSS degradations
(symmetrical and non-symmetrical).

8. Add the tolerance values to the optical and mechanical drawings.

Rule of thumb: If the tolerance calculation doesn't take longer than
all the rest of the design work put together, you probably haven't
done it properly.

PHASE II

1. Phone call from the workshop foreman, "We have a batch of XYs
which are slightly outside tolerance"

2. Phone call from sales manager "This job has to be out of the door
by noon tomorrow, we can't afford to reject them"

3. Imagined conversation with foreman, "If you could accept this
batch, why wasn't that the tolerance you put on the drawing in the
first place?"

4. Imagined conversation with sales manager, "You are jeopardizing all
our jobs, Jonah, we need to throw you overboard to save the ship."

5. Write on concession note something like, "These components are
acceptable provided the spacer XY is made undersize by Z mm/provided
the component PQ has a thickness error of less than R mm/provided the
complete instrument passes the final MTF test" - anything that allows
you to accept the batch, and still maintain your self-respect and be
in control of product quality.

I think the key aspect of tolerances is that the decision-making
process is from day one a dialogue with the workshops (optical and
mechanical), not a harangue.

Brian
Ancient and Modern Optics

Helpful person

Posted: Fri May 02, 2008 7:55 am

Joined: 22 Jun 2004 Posts: 658

On May 2, 5:37 am, Brian <brian4052...@yahoo.co.uk> wrote:

Quote:

On May 1, 7:47 pm, Helpful person <rrl...@yahoo.com> wrote:
snip

Has anyone written a book covering "real world" lens tolerancing?

Not that I know of, but a chap I worked with at Rank/Marconi/BAe
called Alan Stuart did an MSc thesis on it at Imperial College in the
70s.
Brian
Ancient and Modern Optics

There's a name from the past. I haven't seen Alan since he was at
Debden and I was in Edinburgh in the mid seventies.

Helpful person

Posted: Fri May 02, 2008 8:59 am

Joined: 22 Jun 2004 Posts: 658

On May 1, 2:47 pm, Helpful person <rrl...@yahoo.com> wrote:

Quote:

2. What is your nominal and required performance? If your system is
diffraction limited errrors introduced from various elements do not
add up in a linear manner. In other words, if you plot resolution
against wavefront aberration it is not a linear relationship.

Actually they don't add up in a linerar manner for non diffraction
systems either. In general, tolerances are not orthogonal.

Philip McCulloch

Posted: Fri May 02, 2008 10:32 am

Guest

On May 1, 1:16 pm, "Richard F.L.R.Snashall" <rf...@notnotrcn.com>
wrote:

Quote:

Depending on the system opto-mech tolerancing can be pretty straight
forward or pretty painful.

You are asking if a 0.0001 wave change in RMS wavefront is acceptable,
well it completely depends on the system. Say for a visible imaging
system I doubt 0.0001wavefront error would be any more noticeable than
a gnat's fart.

If this is your first attempt to tolerance an optical system I highly
recommend reading up on it in the literature and also learn about
optical testing methods (the tests conducted to verify tolerance
spec) Bob Fischer's book "Optical System Design" dedicates 40 or so
pages to the topic and should serve as a good launching pad. He also
includes rules of thumb to follow that I find helpful identifying the
most costly parameters to tolerance tightly.

Good luck,
philip

Guest

Posted: Fri May 02, 2008 4:06 pm

On May 2, 4:32 pm, Philip McCulloch <philip.mccull...@gmail.com>
wrote:

Quote:

On May 1, 1:16 pm, "Richard F.L.R.Snashall" <rf...@notnotrcn.com
wrote:

Philli...@yahoo.com wrote:
Hi y'all

What is a reasonable allowed change for a tolerance in wave RMS when
tolerance a system in Code V?

Is .0001 out of the question?

Thanks.

Why would a percentage not fit?

Depending on the system opto-mech tolerancing can be pretty straight
forward or pretty painful.

You are asking if a 0.0001 wave change in RMS wavefront is acceptable,
well it completely depends on the system. Say for a visible imaging
system I doubt 0.0001wavefront error would be any more noticeable than
a gnat's fart.

If this is your first attempt to tolerance an optical system I highly
recommend reading up on it in the literature and also learn about
optical testing methods (the tests conducted to verify tolerance
spec) Bob Fischer's book "Optical System Design" dedicates 40 or so
pages to the topic and should serve as a good launching pad. He also
includes rules of thumb to follow that I find helpful identifying the
most costly parameters to tolerance tightly.

Good luck,
philip

Yes tolerancing is really new to me. I was pretty dissapointed when my
amazing design couldn't be manufactured.

Page 1 of 1 All times are GMT - 5 Hours The time now is Mon May 12, 2008 9:36 am Science Forum Index